70 ALASKAN GLACIER STUDIES
clay. The only difficulty in travelling and packing camp outfit over this surface was the unstable position of the rock fragments resting on the glacier, and the necessity of constantly climbing ridges and knobs and descending into pits and valleys, in whatever direction one travelled. There seemed to be no system to the hillocks and depressions, their origin being evidently differential ablation of the under ice which the moraine protected in varying degrees according to the local thickness of the moraine cover.
In the upper portion of this morainic desert the amount of visible ice increased, the moraine cover became notably thinneri and the surface was less irregular. Some cre-vassing also appeared, but nowhere in sufficient amount to prevent travelling over the glacier except in one small section where a dome of crevassed ice appeared in the midst of the moraine desert. It was so abnormal a feature that we photographed it, thinking that it represented the updoming of ice in passing over some elevation in the valley bottom. We predicted the existence here of an unborn nunatak, but the crevassed dome is now interpreted as the first sign of the coming transformation of the glacier which was observed in full progress in 1906. Still farther up the glacier, clear ice predominated and finally the moraine entirely disappeared, but we did not visit this part of the glacier and, therefore, cannot describe it in detail.
Nowhere in the upper part of the valley portion of Atrevida Glacier did vegetation grow in the ablation moraine, for it was shifting rapidly. But farther down, near the end of the valley portion, scattered alder bushes and clusters of alder grew, extending farther up the valley on the eastern margin than in the center, because the lateral moraine deposits were thicker and, therefore, under more stable conditions. Just outside the mountain valley portion of the glacier the ablation moraine was covered with alder, and farther out, with, cottonwood and spruce. These bushes and trees were mature, indicating a long condition of stagnation in this part of the glacier. The stagnant outer part of the Atrevida was a true piedmont bulb, extending four or five miles beyond the mountain front and expanding fan-shaped until it attained a width of about twice that of the valley portion. On the western side it coalesced with Lucia Glacier. As already stated in the description of that glacier, the two glaciers were separated in the lower p6rtion by a V-shaped valley, then by an interlobate section on which a narrow strip of alder grew, and above this the two glaciers again separated a short distance below Terrace Point, which, with the mountain back of it form the dividing walls between the valley portions of the two glaciers.
Both at Terrace Point and along the eastern side of the valley portion of Atrevida Glacier the margin had a thickly moraine^covered, moderately-sloping embankment' which we could ascend at any point. It formed one wall of a marginal valley, carrying some drainage. On the west side there was little drainage from either the Lucia or Atrevida, and only a small stream escaped in the V-shaped valley between the interlobate portion of the coalesced Lucia and Atrevida; and there were only small streams emerging from the outer portion of the piedmont bulb which rose above the Kwik valley. A small stream emerged from the forest-covered part of the piedmont bulb on the eastern side and flowed into Yakutat Bay, but it was easily forded. The main drainage of the glacier emerged from the east side at the base of Amphitheatre Knob. At the point of emergence of this stream, which Hussell called Esker Stream, there was a low cave with vertical or overhanging walls above which the ice edge had a slope of about 35°, down which a few stones were sliding. The ice was nowhere heavily debris-laden, though dirtiest